Review ArticleAccuracy Evaluation of Circular RNA in Diagnosing LungCancer in a Chinese Population

Zhihao Xiao , Xinglei Chen , Xiaodan Lu, Xuexin Zhong , and Yihui Ling

Institute for Chemical Carcinogenesis, Guangzhou Medical University, Guangzhou, China

Correspondence should be addressed to Yihui Ling; ling_yh@163.com

Received 13 May 2019; Revised 17 August 2019; Accepted 5 September 2019; Published 20 October 2019

Academic Editor: Chiara Nicolazzo

Copyright © 2019 Zhihao Xiao et al. This is an open access article distributed under the Creative Commons Attribution License,which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

Circular RNA (circRNA) is a class of recently discovered noncoding RNA. circRNAs can be used as a potent noninvasive biologicalmarker of cancer owing to their varying expression levels among different cancers. This meta-analysis was performed to assess theaccuracy of circRNAs in diagnosing lung cancer. A total of eight studies identified through searching the PubMed, Web of Science,Cochrane Library, and Embase from inception to March 20, 2019 were eligible for this meta-analysis. The pooled sensitivity,specificity, positive likelihood ratios, negative likelihood ratios, and diagnostic odds ratio were 0.77 (95% confidence interval(CI): 0.73–0.80; I2 = 8:98%), 0.76 (95% CI: 0.69–0.82; I2 = 63:12%), 3.17 (95% CI: 2.43–4.14; I2 = 33:18%), 0.31 (95% CI: 0.26–0.37; I2 = 20:36%), and 10.26 (95% CI: 6.87–15.31; I2 = 97:18%), respectively. The area under the receiver operatingcharacteristic curve was 0.78 (95% CI: 0.74–0.81). The study confirmed the use of circRNAs in diagnosing lung cancer in aChinese population.

1. Introduction

Circular RNA (circRNA) is a circular single-strand RNA firstdiscovered in plants [1, 2] and abundant in human cells. Insome cases, the abundance of circRNAs exceeds that of asso-ciated linear mRNAs by more than tenfold [3]. CircRNAslack free ends, 5′ cap, and 3′ poly(A) tail and are more stablethan linear RNAs with their ends joining in a circle via phos-phodiester bonds [4]. CircRNAs have many functionsincluding regulating gene transcription and translation viabinding to miRNAs, interacting with proteins, and beingdirectly translated [5]. Recent studies suggested significantadvantages of circRNAs in diagnosing cancers owing to theirprevalence, stability, specificity, and conservatism [6].

Experiments proved that circRNAs were associated withproliferation, apoptosis, invasion, and migration of tumorcells, and the expression varied with tumor cells, thus helpingin diagnosing and predicting the prognosis of tumors. Li et al.[7] found that Hsa_circ_0000096 was significantly downreg-ulated in gastric cancer tissues and cell lines and was associ-ated with the invasion and staging of tumors. Ma et al. [8]

reported that circRNA-000284 was significantly higher incervical cancer cells than in cervical epithelial cells and couldserve as a biological marker. Li et al. [9] considered cir-cHIPK3 as the novel therapeutic target of bladder carcinoma.Some meta-analyses summarizing the role of circRNAs ondiagnosis and prognosis were reported [10–13]. Huanget al. [10] investigated the prognostic and diagnostic signifi-cance of the expression of circRNAs in patients with hepato-cellular carcinoma. Wang et al. [11], Li et al. [12], and Ding[13] performed a meta-analysis on the value of circRNAs asa biological marker of tumors. However, they just exploredthe role of circRNAs in diagnosing tumors but not speciallyor independently diagnosing lung cancer.

Lung cancer accounts for 25% of cancer-related deathsworldwide, which is much higher compared with other can-cers such as breast carcinoma, prostatic carcinoma, and colo-rectal carcinoma. The survival rate of patients with lungcancer is relatively low, possibly due to the lack of early test-ing methods [14]. The role of circRNAs in lung cancer wasfirst reported in 2018 by Qu et al. [5] who found that hsa_circ_00013958 promoted the proliferation, invasion, and

HindawiDisease MarkersVolume 2019, Article ID 7485389, 8 pageshttps://doi.org/10.1155/2019/7485389

metastasis of lung adenocarcinoma cells. A large number ofrecent studies explored the relationship between circRNAsand lung cancer [15–21].

2. Materials and Methods

2.1. Literature Search. Electronic databases, includingPubMed, Web of Science, Cochrane Library, and Embase,were systematically searched from inception to March 20,2019. The following search terms were used: circRNA, circu-lar RNA, lung cancer, lung neoplasm, lung adenocarcinoma,non-small-cell lung cancer, NSCLC, pulmonary cancer, andpulmonary neoplasm. In addition, the reference lists of eligi-ble studies were manually searched to guarantee the compre-hensiveness of the literature.

2.2. Inclusion and Exclusion Criteria. The inclusion criteriawere as follows: (1) studies analyzing the relationshipbetween circRNAs and lung cancer, (2) studies providingdata on the sensitivity and specificity, and (3) studies involv-ing ≥30 patients and controls. The exclusion criteria were asfollows: (1) repetitive research; (2) letters, editorials, com-mentaries, or abstracts; (3) studies involving ineligiblepatients or controls; (4) studies lacking data; or (5) studiesin a non-English language. If the results came from the over-lapping population, only the first study or the most completestudy was included.

2.3. Data Extraction and Quality Assessment. Two reviewers(Xuexin Zhong and Zhihao Xiao) extracted data. The dis-crepancies were resolved by the third reviewer (XiaodanLu) if needed. The following information was extracted fromeach study: first author name, year of publication, country,sample size, sample type, sensitivity, specificity, area underthe receiver operating characteristic (ROC) curve (AUC),

testing methods, tumor staging, reference gene, and differen-tial expression of circRNAs.

2.4. Statistical Analysis. Stata 12.0 software and Meta-DiSc1.4 were used for statistical analysis. The sensitivity, specific-ity, positive likelihood ratio (PLR), negative likelihood ratio(NLR), diagnostic odds ratio (DOR), 95% confidential inter-vals (95% CI), summary ROC (SROC) curve, and AUC werecalculated for quality assessment. The significance level wasset as P < 0:05. The heterogeneity induced by the thresholdeffect of included studies was tested using the Spearman cor-relation analysis. Cochran’sQ and I2 tests were used to assessthe heterogeneity of data. I2 > 50% indicated significant het-erogeneity. A subgroup analysis was performed based onsample type, cancer type, reference gene type, and differentialexpression of circRNAs. The potential source of heterogene-ity by the nonthreshold effect was analyzed by regressionanalysis. A Fagan nomogram was used to calculate the post-test probabilities. Finally, publication bias was assessed.

3. Results

3.1. Search Results. The flowchart of the study selection pro-cess is shown in Figure 1. The review of the literature identi-fied 388 studies, of which 209 repeated ones were excludedand the remaining 179 ones were screened based on titlesand abstracts. Subsequently, 165 studies, including reviews,letters, conference abstracts, and 2 studies on hsa_circ_0102533 [22] and circFARSA [23], were further excluded,owing to the inability of constructing a 2 × 2 contingencytable. A total of 8 studies [15–21, 24] and 10 eligible studies,involving 668 patients with lung cancer and 153 healthy con-trols, were assessed during this meta-analysis. The character-istics of eight studies are shown in Table 1. Surprisingly, alleight studies identified using search terms, inclusion criteria,and exclusion criteria were performed in China.

Iden

tifica

tion

Scre

enin

gEl

igib

ility

Incl

uded

Records a�er removing duplicates(n = 179)

Full-text studiesexcluded:

Studies with insufficient data to construct a 2 ⨯ 2 table Studies included in quantitative synthesis

(meta-analysis)

Records excluded:

Reviews, letters, conference abstracts, and

irrelevant studies with respect to circRNAs or lung malignancy Full-text articles assessed for eligibility

(n = 10)

Records identified throughdatabase search (n = 387)

Additional records identifiedthrough other sources (n = 1)

Figure 1: Flow chart of the study selection process.

2 Disease Markers

Table1:Characteristics

ofeightstud

iesinclud

edin

themeta.

Autho

rYear

circRNAs

Cou

ntry

Sampletype

No.of

cases

No.of

controls

AUC

Sensitivity

Specificity

Testing

metho

dTNM

staging

circRNA

expression

References

I–II

III–IV

Zhu

etal.

2017

hsa_circ_0013958

China

Blood

3030

0.794

0.667

0.933

qRT-PCR

2821

Upregulated

[24]

Lietal.

2018

hsa_circ_0079530

China

Tissue

9292

0.756

0.762

0.721

qRT-PCR

——

Upregulated

[15]

Zon

getal.

2018

circRNAs_102231

China

Tissue

5757

0.897

0.812

0.887

qRT-PCR

3027

Upregulated

[16]

Zhang

etal.2018

hsa_circ_0014130

China

Tissue

4646

0.878

0.87

0.848

qRT-PCR

3610

Upregulated

[17]

Zhang

etal.2018

circ_F

OXO3

China

Tissue

4545

0.782

0.8

0.733

qRT-PCR

——

Dow

nregulated

[18]

Lietal.

2018

circ_P

VT1

China

Blood

4545

0.794

0.711

0.8

qRT-PCR

3137

Upregulated

[19]

China

Tissue

6868

0.803

0.825

0.675

qRT-PCR

3137

Upregulated

Chenetal.

2019

circRNAs_100146

China

Tissue

4040

0.643

0.725

0.575

qRT-PCR

——

Upregulated

[20]

Liuetal.

2019

hsa_circ_0005962

China

Blood

153

540.73

0.719

0.7222

qRT-PCR

9655

Upregulated

[21]

hsa_circ_0086414

China

Blood

153

540.78

0.7712

0.6667

qRT-PCR

9655

Dow

nregulated

Note:data

notextracted;

AUC:areaun

derthereceiver

operatingcharacteristiccurve.

3Disease Markers

Study ID Specificity (95% CI)Specificity (95% CI)

Specificity

0.77 [0.70 - 0.84] 0.67 [0.53 - 0.79]

0.68 [0.55 - 0.78]

0.93 [0.78 - 0.99]

0.72 [0.58 - 0.84]

0.57 [0.41 - 0.73]

0.80 [0.65 - 0.90]

0.73 [0.58 - 0.85]

0.85 [0.71 - 0.94]

0.88 [0.76 - 0.95]

0.72 [0.61 - 0.81]

0.76 [0.69 - 0.82]

Q = 24.40, df = 9.00, P = 0.00

I2 = 63.12 [37.94 - 88.30]

0.82 [0.71 - 0.91]

0.67 [0.47 - 0.83]

0.72 [0.64 - 0.79]

0.73 [0.56 - 0.85]

0.71 [0.56 - 0.84]

0.80 [0.65 - 0.90]

0.87 [0.74 - 0.95]

0.81 [0.68 - 0.90]

0.76 [0.66 - 0.84]

0.77 [0.73 - 0.80]

Q = 9.89, df = 9.00, P = 0.36

I2 = 8.98 [0.00 - 97.95]

Study ID

Liu et al/2019

Liu et al/2019

Chen et al/2019

Li et al/2018

Zhang et al/2018

Zhang et al/2018

Zong et al/2018

Li et al/2018

Combined

0.5 1.0 1.00.4

Sensitivity

Li et al/2018

Zhu et al/2017

Liu et al/2019

Liu et al/2019

Chen et al/2019

Li et al/2018

Zhang et al/2018

Zhang et al/2018

Zong et al/2018

Li et al/2018

Combined

Li et al/2018

Zhu et al/2017

Figure 2: Forest plots of the sensitivity and specificity for circRNAs in diagnosing lung cancer.

2.31 [1.57 - 3.41] 0.34 [0.24 - 0.49]

0.26 [0.15 - 0.45]

0.36 [0.21 - 0.60]

0.39 [0.29 - 0.53]

0.48 [0.27 - 0.85]

0.38 [0.22 - 0.58]

0.27 [0.15 - 0.50]

0.15 [0.07 - 0.33]

0.22 [0.13 - 0.38]

0.33 [0.23 - 0.49]

0.31 [0.26 - 0.37]

Q = 11.30, df = 9.00, P = 0.26

I2 = 20.36 [0.00 - 76.67]

2.55 [1.77 - 3.65]

10.00 [2.56 - 39.06]

2.59 [1.66 - 4.02]

1.71 [1.13 - 2.56]

3.56 [1.93 - 6.57]

3.00 [1.81 - 4.98]

5.71 [2.86 - 11.41]

6.57 [3.25 - 13.30]

2.69 [1.91 - 3.80]

3.17 [2.43 - 4.14]

Q = 22.80, df = 9.00, P = 0.01

I2 = 33.18 [33.18 - 87.87]

1.1 39.1

DLR positive

0 1

DLR negative

Study ID DLR positive (95% CI) Study ID DLR negative (95% CI)

Liu et al/2019

Liu et al/2019

Chen et al/2019

Li et al/2018

Zhang et al/2018

Zhang et al/2018

Zong et al/2018

Li et al/2018

Combined

Li et al/2018

Zhu et al/2017

Liu et al/2019

Liu et al/2019

Chen et al/2019

Li et al/2018

Zhang et al/2018

Zhang et al/2018

Zong et al/2018

Li et al/2018

Combined

Li et al/2018

Zhu et al/2017

Figure 3: Forest plots of the positive likelihood ratio and the negative likelihood ratio for circRNAs in diagnosing lung cancer.

4 Disease Markers

3.2. Threshold Effect. The threshold effect was evaluatedwith the Spearman rank correlation. The Spearman corre-lation coefficient was 0.079 (P = 0:829), suggesting nothreshold effect.

3.3. Results of Meta-Analysis. Significant heterogeneity wasassessed using the random-effects model (I2 > 50%). For thevalue of circRNAs in diagnosing lung cancer, the pooled sen-sitivity, specificity, PLR, NLR, and DOR were 0.77 (95% CI:0.73–0.80; I2 = 8:98), 0.76 (95% CI: 0.69–0.82; I2 = 63:12%),3.17 (95% CI: 2.43–4.14; I2 = 33:18%), 0.31 (95% CI: 0.26–0.37; I2 = 20:36%), and 10.26 (95% CI: 6.87–15.31; I2 =97:18%), respectively. AUC was 0.78 (95% CI: 0.74–0.81).Forest plots and SROC are shown in Figures 2–5. Fagan’snomogram is shown in Figure 6. If the pretest probabilitywas 20%, the posttest probability increased to 44%. The pre-test likelihood ratio (LR) was 3%, and the posttest LRdecreased to 7%. The NLR was 0.31. An LR scattergramwas used to evaluate the clinical value of different diagnostictests and divided into four quadrants (Figure 7). All 10 eligi-ble studies were in the right lower quadrants, suggesting thatcircRNAs were useful in diagnosing lung cancer.

3.4. Regression Analysis. The I2 value was 96.84%, suggestingsignificant heterogeneity. The sample type, cancer type, refer-

ence gene type, and differential expression of circRNAs weretaken as potential causes of heterogeneity, and a metaregres-sion analysis was performed. No significant causes for het-erogeneity were found. The results are shown in Table 2.

3.5. Subgroup Analysis. A subgroup analysis was performedbased on the sample type, sample size, cancer type, referencegene type, and differential expression of circRNAs. Althoughthe sample size did not contribute to heterogeneity, the sen-sitivity, specificity, and DOR of blood samples were 0.72,0.78, and 9.32, while the corresponding values of tissue sam-ples were 0.80, 0.75, and 11.67, respectively. These findingssuggested that tissue circRNAs were slightly superior toblood circRNAs in diagnosis. The sensitivity, specificity,and DOR of the non-small-cell carcinoma subgroup were0.78, 0.70, and 9.66, while the corresponding values of thelung adenocarcinoma subgroup were 0.75, 0.80, and 12.54,respectively, suggesting the superiority of circRNAs in lungadenocarcinoma over those in non-small-cell carcinoma.The subgroup analysis of the reference gene was not con-ducted because only one study considered β-actin as the ref-erence gene. The results are shown in Table 3.

3.6. Publication Bias. The publication bias of the includedstudies was tested using Deeks’ funnel plot, and significantdifferences in the slope rate (P < 0:05) suggested the publica-tion bias. The funnel plot was constructed with the Stata 12.0software (Figure 8), and the results showed no publicationbias (P = 0:24).

Study ID

6.74 [3.42 - 13.31]

9.76 [4.37 - 21.81]

28.00 [5.52 - 141.91]

6.65 [3.33 - 13.29]

3.57 [1.40 - 9.09]

9.85 [3.72 - 26.08]

11.00 [4.11 - 29.45]

37.14 [11.46 - 120.42]

29.87 [10.68 - 83.56]

8.08 [4.18 - 15.63]

10.26 [6.87 - 15.31]

Q = 313.60, df = 9.00, P = 0.00

I2 = 97.18 [96.24 - 98.11]

Liu et al/2019

Liu et al/2019

Chen et al/2019

Li et al/2018

Zhang et al/2018

Zhang et al/2018

Zong et al/2018

Li et al/2018

Combined

1 142

Odds ratio

Li et al/2018

Zhu et al/2017

Odds ratio (95% CI)

Figure 4: Forest plots of the diagnostic odds ratio for circRNAs indiagnosing lung cancer.

SROC with prediction & confidence contours1.0

0.5

0.01.0 0.5 0.0

Observed dataSummary operating pointSENS = 0.77 [0.73 - 0.80]SPEC = 0.76 [0.69 - 0.82]SROC curveAUC = 0.78 [0.74 - 0.81]

95% confidence contour95% prediction contour

Specificity

Sens

itivi

ty

3

2 4 910

758

61

Figure 5: Summary receiver operating characteristic curve forcircRNAs in diagnosing lung cancer.

5Disease Markers

4. Discussion

The present meta-analysis enrolled 8 studies from 2017 to2019 and systemically reviewed 10 circRNAs in diagnosinglung cancer. The results showed that the AUC was 0.78,and the pooled sensitivity, specificity, and DOR were, respec-tively, 0.77 (95% CI: 0.73–0.80), 0.76 (95% CI: 0.69–0.82),and 10.26 (95% CI: 6.87–15.31). The findings suggested thediagnostic value of circRNAs for lung cancer. The includedstudies involved only a preliminary analysis of the role ofone or two circRNAs in diagnosing lung cancer, with smallsample size and sensitivity varying from 0.511 to 0.884(lower than pooled sensitivity in six studies), specificity from0.575 to 0.933 (lower than pooled specificity in six studies),and AUC from 0.643 to 0.897 (lower than or equal toAUC in this meta-analysis in five studies). The sensitivity,specificity, and AUC fluctuated largely among these studies,possibly due to the involvement of one or two circRNAs andsmall sample size.

The Spearman correlation coefficient was calculated totest the threshold effect and was 0.79 (P = 0:829), suggestingthat the threshold effect was not the cause of heterogeneity.In addition, the regression analysis of sample type, lungcancer type, reference type, and differential expression ofcircRNAs showed that these factors did not cause heteroge-neity. The causes of heterogeneity could not be directly foundusing the studies included in this analysis. Neither study was

a randomized controlled study, possibly leading to heteroge-neity. However, this hypothesis should be further verified.Additionally, the present meta-analysis included 8 studiesand systematically reviewed the value of 10 different cir-cRNAs in diagnosing lung cancer. The analysis revealed thatthe expression levels of these circRNAs were different, whichmight be one of the sources of heterogeneity and also a com-mon problem in this type of analysis.

In conclusion, this systematic review of data extractedfrom eight studies showed the value of circRNAs in diagnos-ing lung cancer. These studies were primarily based on thetesting of lung cancer tissues. Blood and exocrine secretions

100

LUQ: exclusion & confirmationLRP>10, LRN<0.1RUQ: confirmation onlyLRP>10, LRN>0.1LLQ: exclusion onlyLRP<10. LRN<0.1RLQ: no exclusion or confirmationLRP<10, LRN>0.1

with 95% confidence intervalsSummary LRP & LRN for index test

10

11

Negative likelihood ratio

2

9110

7

9

6

4

8

3

Figure 7: A likelihood ratio scattergram.

Table 2: Relative diagnostic odds ratio of covariants in themetaregression analysis.

RDOR 95% CI P value

Sample type 2.07 (0.62–6.95) 0.1710

Cancer type 1.38 (0.41–4.67) 0.5064

Reference 0.27 (0.04–1.63) 0.1125

Expression of circRNAs 0.81 (0.21–3.18) 0.6918

0.1

0.20.30.50.7

1

2357

Likelihood ratio

10005002001005020105210.50.20.10.050.020.010.0050.0020.001

10

203040506070

Pre-

test

prob

abili

ty (%

)

Post-

test

prob

abili

ty (%

)

80

909395979899

99.399.599.799.899.9 0.1

Prior prob (%) = 20LR_Positive = 3Post_Prob_Pos (%) = 44LR_Negative = 0.31Post_Prob_Neg (%) = 7

0.20.30.50.71

235710

20304050607080

90939597989999.399.599.799.8

99.9

Figure 6: Fagan’s nomogram for likelihood ratios.

6 Disease Markers

were less used, and clinical data were limited. Therefore, therelationship of circRNAs in the blood or exocrine secretionswith lung cancer needs further investigation, and the findingmight help in the development of molecular markers of diag-nosis and prognosis. Further, the recent studies explored therole of a single circRNA in diagnosing and treating cancerusing a small sample size; no multicenter and large-samplestudies were reported. Hence, the diagnostic accuracy andstability of circRNAs need further elucidation.

Conflicts of Interest

The authors declared that they have no conflict of interests.

Acknowledgments

This study was supported by Guangdong MedicalResearch Fund Project (A2018164) and Guangdong Prov-ince General University Characteristic Innovation Project(2018KTSCX185).

References

[1] S. Memczak, M. Jens, A. Elefsinioti et al., “Circular RNAs are alarge class of animal RNAs with regulatory potency,” Nature,vol. 495, no. 7441, pp. 333–338, 2013.

[2] H. L. Sanger, G. Klotz, D. Riesner, H. J. Gross, and A. K.Kleinschmidt, “Viroids are single-stranded covalently closedcircular RNA molecules existing as highly base-paired rod-likestructures,” Proceedings of the National Academy of Sciences ofthe United States of America, vol. 73, no. 11, pp. 3852–3856,1976.

[3] W. R. Jeck, J. A. Sorrentino, K. Wang et al., “Circular RNAs areabundant, conserved, and associated with ALU repeats,” RNA,vol. 19, no. 2, pp. 141–157, 2013.

[4] X. O. Zhang, H. B. Wang, Y. Zhang, X. Lu, L. L. Chen, andL. Yang, “Complementary sequence-mediated exon circulari-zation,” Cell, vol. 159, no. 1, pp. 134–147, 2014.

[5] S. Qu, Z. Liu, X. Yang et al., “The emerging functions and rolesof circular RNAs in cancer,” Cancer Letters, vol. 414, pp. 301–309, 2018.

[6] X. Wang and L. Fang, “Advances in circular RNAs and theirroles in breast cancer,” Journal of Experimental & ClinicalCancer Research, vol. 37, no. 1, p. 206, 2018.

[7] P. Li, H. Chen, S. Chen et al., “Circular RNA 0000096 affectscell growth and migration in gastric cancer,” British Journalof Cancer, vol. 116, no. 5, pp. 626–633, 2017.

[8] H. B. Ma, Y. N. Yao, J. J. Yu, X. X. Chen, and H. F. Li, “Exten-sive profiling of circular RNAs and the potential regulatoryrole of circRNA-000284 in cell proliferation and invasion ofcervical cancer via sponging miR-506,” American Journal ofTranslational Research, vol. 10, no. 2, pp. 592–604, 2018.

[9] Y. Li, F. Zheng, X. Xiao et al., “CircHIPK3 sponges miR-558 tosuppress heparanase expression in bladder cancer cells,”EMBO reports, vol. 18, no. 9, pp. 1646–1659, 2017.

Table 3: Summary results of the subgroup analysis for circRNAs in diagnosing lung cancer.

Number ofstudies

Sensitivity(95% CI)

Specificity(95% CI)

PLR (95% CI) NLR (95% CI) DOR (95% CI) AUC

Sample size

Blood 4 0.72 (0.66–0.78) 0.78 (0.66–0.87) 3.30 (2.12–5.15) 0.35 (0.29–0.43) 9.32 (5.35–16.23) 0.78

Tissue 6 0.80 (0.75–0.84) 0.75 (0.66–0.82) 3.16 (2.22–4.49) 0.27 (0.20–0.36) 11.67 (6.36–21.39) 0.84

Lung cancer

Non-small-celllung cancer

6 0.78 (0.73–0.83) 0.70 (0.66–0.79) 2.87 (2.22–3.72) 0.30 (0.23–0.38) 9.66 (6.00–15.55) 0.82

Lung adenocarcinoma 4 0.75 (0.69–0.79) 0.81 (0.68–0.90) 3.93 (2.20–7.03) 0.31 (0.25–0.40) 12.54 (5.96–26.39) 0.76

Reference gene

GAPDH 9 0.76 (0.72–0.79) 0.75 (0.68–0.81) 3.00 (2.31–3.90) 0.33 (0.28–0.38) 9.41 (6.66–13.31) 0.77

circRNA expressionupregulation

8 0.76 (0.72–0.80) 0.77 (0.69–0.84) 3.38 (2.39–4.77) 0.31 (0.25–0.38) 10.96 (6.57–18.29) 0.79

Note: PLR: positive likelihood ratio; NLR: negative likelihood ratio; AUC: area under the receiver operating characteristic curve.

6

5 4 3

8

2

9

1

10

.06

Deek’s funnel plot asymmetry testP value = 0.24

StudyRegressionline

.08

.1

.12

.141 10 100 1000

Diagnostic odds ratio

1/ro

ot (E

SS)

Figure 8: Deeks’ funnel plot evaluating the potential publicationbias of the included studies.

7Disease Markers

[10] X. Huang, W. Zhang, and Z. Shao, “Prognostic and diag-nostic significance of circRNAs expression in hepatocellularcarcinoma patients: a meta‐analysis,” Cancer Medicine, vol. 8,no. 3, pp. 1148–1156, 2019.

[11] M. Wang, Y. Yang, J. Xu, W. Bai, X. Ren, and H. Wu,“CircRNAs as biomarkers of cancer: a meta-analysis,” BMCCancer, vol. 18, no. 1, p. 303, 2018.

[12] Y. Li, X. Zeng, J. He et al., “Circular RNA as a biomarker forcancer: a systematic meta-analysis,” Oncology Letters, vol. 16,no. 3, pp. 4078–4084, 2018.

[13] H. X. Ding, Z. Lv, Y. Yuan, and Q. Xu, “The expression of cir-cRNAs as a promising biomarker in the diagnosis and progno-sis of human cancers: a systematic review and meta-analysis,”Oncotarget, vol. 9, no. 14, pp. 11824–11836, 2018.

[14] I. Balgkouranidou, T. Liloglou, and E. S. Lianidou, “Lung can-cer epigenetics: emerging biomarkers,” Biomarkers in Medi-cine, vol. 7, no. 1, pp. 49–58, 2013.

[15] J. Li, J. Wang, Z. Chen, Y. Chen, and M. Jin, “Hsa_circ_0079530 promotes cell proliferation and invasion in non-small cell lung cancer,” Gene, vol. 665, pp. 1–5, 2018.

[16] L. Zong, Q. Sun, H. Zhang et al., “Increased expression of cir-cRNA_102231 in lung cancer and its clinical significance,”Biomedicine & Pharmacotherapy, vol. 102, pp. 639–644, 2018.

[17] S. Zhang, X. Zeng, T. Ding et al., “Microarray profile of circularRNAs identifies hsa_circ_0014130 as a new circular RNA bio-marker in non-small cell lung cancer,” Scientific Reports, vol. 8,no. 1, p. 2878, 2018.

[18] Y. Zhang, H. Zhao, and L. Zhang, “Identification of the tumor-suppressive function of circular RNA FOXO3 in non-small celllung cancer through sponging miR-155,” Molecular MedicineReports, vol. 17, no. 6, pp. 7692–7700, 2018.

[19] X. Li, Z. Zhang, H. Jiang et al., “Circular RNA circPVT1 pro-motes proliferation and invasion through sponging miR-125band activating E2F2 signaling in non-small cell lung cancer,”Cellular Physiology and Biochemistry, vol. 51, no. 5, pp. 2324–2340, 2018.

[20] L. Chen, A. Nan, N. Zhang et al., “Circular RNA 100146 func-tions as an oncogene through direct binding to miR-361-3pand miR-615-5p in non-small cell lung cancer,” MolecularCancer, vol. 18, no. 1, p. 13, 2019.

[21] X. X. Liu, Y. E. Yang, X. Liu et al., “A two-circular RNA signa-ture as a noninvasive diagnostic biomarker for lung adenocar-cinoma,” Journal of Translational Medicine, vol. 17, no. 1,p. 50, 2019.

[22] X. Zhou, H. Y. Liu, W. Y. Wang, H. Zhao, and T. Wang, “Hsa_circ_0102533 serves as a blood-based biomarker for non-small-cell lung cancer diagnosis and regulates apoptosisin vitro,” International Journal of Experimental Pathology,vol. 11, no. 9, pp. 4395–4404, 2018.

[23] D. Hang, J. Zhou, N. Qin et al., “A novel plasma circular RNAcircFARSA is a potential biomarker for non-small cell lungcancer,” Cancer Medicine, vol. 7, no. 6, pp. 2783–2791, 2018.

[24] X. Zhu, X. Wang, S. Wei et al., “hsa_circ_0013958: a circularRNA and potential novel biomarker for lung adenocarci-noma,” The FEBS Journal, vol. 284, no. 14, pp. 2170–2182,2017.

8 Disease Markers

Stem Cells International

Hindawiwww.hindawi.com Volume 2018

Hindawiwww.hindawi.com Volume 2018

MEDIATORSINFLAMMATION

of

EndocrinologyInternational Journal of

Hindawiwww.hindawi.com Volume 2018

Hindawiwww.hindawi.com Volume 2018

Disease Markers

Hindawiwww.hindawi.com Volume 2018

BioMed Research International

OncologyJournal of

Hindawiwww.hindawi.com Volume 2013

Hindawiwww.hindawi.com Volume 2018

Oxidative Medicine and Cellular Longevity

Hindawiwww.hindawi.com Volume 2018

PPAR Research

Hindawi Publishing Corporation http://www.hindawi.com Volume 2013Hindawiwww.hindawi.com

The Scientific World Journal

Volume 2018

Immunology ResearchHindawiwww.hindawi.com Volume 2018

Journal of

ObesityJournal of

Hindawiwww.hindawi.com Volume 2018

Hindawiwww.hindawi.com Volume 2018

Computational and Mathematical Methods in Medicine

Hindawiwww.hindawi.com Volume 2018

Behavioural Neurology

OphthalmologyJournal of

Hindawiwww.hindawi.com Volume 2018

Diabetes ResearchJournal of

Hindawiwww.hindawi.com Volume 2018

Hindawiwww.hindawi.com Volume 2018

Research and TreatmentAIDS

Hindawiwww.hindawi.com Volume 2018

Gastroenterology Research and Practice

Hindawiwww.hindawi.com Volume 2018

Parkinson’s Disease

Evidence-Based Complementary andAlternative Medicine

Volume 2018Hindawiwww.hindawi.com

Submit your manuscripts atwww.hindawi.com